RNM, I may well be interested in parting with the VP-7725, not immediately, but possibly in the not to distant future. However I do need something that is "good enough" somewhere on the bench... thus the interest in the QA400.
The issue is that I apparently over voltaged the input of one of the two channels. It is spec'd on the front panel at "25v" on one of the two balanced inputs and "160v" on the other. Of course I used the lower voltage one. Of course. That's because I am highly qualified to run exotic unrepairable test gear. Highly qualified.
I'm certainly open to the 7722 or whatever model schematic too. Send it this way.
Hopefully it does have similar pedigree, but this one is all uproc controlled and the switches are all soft switches, and there is some auto ranging going on too...
Demian, if I had spare time, I'd ship it out to you and fly out to join the party.
Of course you'd learn just how little I know...
The issue is that I apparently over voltaged the input of one of the two channels. It is spec'd on the front panel at "25v" on one of the two balanced inputs and "160v" on the other. Of course I used the lower voltage one. Of course. That's because I am highly qualified to run exotic unrepairable test gear. Highly qualified.
I'm certainly open to the 7722 or whatever model schematic too. Send it this way.
Hopefully it does have similar pedigree, but this one is all uproc controlled and the switches are all soft switches, and there is some auto ranging going on too...
Demian, if I had spare time, I'd ship it out to you and fly out to join the party.
Of course you'd learn just how little I know...Attachments
Demian, if I had spare time, I'd ship it out to you and fly out to join the party.
If you want to ship it my way I'll be happy to explore it and maybe fix it. No promises or time frame commitments but it would be interesting to check out.
Tektronix will also repair it and calibrate it in a timely manner. Call them (West Coast Time). They are not expensive, I found out.
I think the 7725 is the better unit, BTW. But, its your call.
I have test equipment coming out my ears... but the 7725A is one worth my buying.
THx-RNMarsh
I think the 7725 is the better unit, BTW. But, its your call.
I have test equipment coming out my ears... but the 7725A is one worth my buying.
THx-RNMarsh
Last edited:
Well thanks much!
Let's see how things progress, I'm in no position to really deal with much right now, the best I can do is to essentially snipe online interleaved between real world responsibilities.
I think the newer unit, as sold (allegedly sold) by Levear has slightly better specs.
_-_-
Let's see how things progress, I'm in no position to really deal with much right now, the best I can do is to essentially snipe online interleaved between real world responsibilities.
I think the newer unit, as sold (allegedly sold) by Levear has slightly better specs.
_-_-
725B Service Manual On CDR...
I just got notification of this Ebay listing.
I can't get posting a link here to work so search Ebay for "Operating manual with schematics for ShibaSoku 725B".
CDR for US$40.00 plus postage.
This seller also has "Service manual with schematics for National VP-7722A"
Dan.
I just got notification of this Ebay listing.
I can't get posting a link here to work so search Ebay for "Operating manual with schematics for ShibaSoku 725B".
CDR for US$40.00 plus postage.
This seller also has "Service manual with schematics for National VP-7722A"
Dan.
Last edited:
I found a link for the manual here with schemtics.
http://kenrockwell.com/audio/panasonic/images/vp-7721a/vp-7721a-manual.pdf
http://kenrockwell.com/audio/panasonic/images/vp-7721a/vp-7721a-manual.pdf
The ADC board in the ShibaSoku AD725C that I have is silkscreened AD725C A14 S0000092-2B. It's fitted with a Burr-Brown ACD80AG-12 which is a 12 bit successive-approximation analog-to digital converter. The board uses the ADC internal clock and the Serial Out connected to an Optocoupler. The ADC input is fed from a Burr-Brown SHC5320 sample and hold.
The board also includes an Analog Devices AD7541ABQ 12 bit multiplying digital to analog converter though I don't know what is used for because I haven't traced out the complete circuit.
According to the patent the input is converted to digital. Noise reduction is performed and the output data is re-clocked at a fixed frequency. This allows fixed frequency filters to be used. Kind of like a heterodyne process.
Is that one of the processes? I didn't get that from the patent.
I know there is a two stage twin or other notch type post reconstruction followed by a tunable BP for singling out harmonics. There is also mention of a comb filter.
Last edited:
The patent is not very clear. There are three cascaded relay tuned notch filters that add up to about -140 dB. There is no servo null. The output of the filters is fed into the ADC. The multiplier probably does an AGC on the input of the ADC. The sample and hold is used for some synchronous sampling tricks and probably uses undersampling to deal with the 100 KHz stuff. The synchronous sampling will tend to reject asynchronous signals (noise) when averaged. The sampled data is stored in probably a local ring memory the patent gets pretty murky here and the actual implementation is a little different I suspect). The memory is read out at that odd sample rate (450 Hz fundamental) and then into the fixed tuned harmonic filters. they use LC filters for narrow bandwidth.
In a new design the complexity would be replaced with a more straightforward ADC and fft. The tuning gets complicated with three stages and all the relays for fixed tuning with that much resolution. If it could be accomplished with the Maxim or JRC "pots" it might be something practical for a DIY project. otherwise the relay cost alone makes it impractical at best.
I looked at the Panasonic input amps and they are quite similar to the Shibasoku. JFET diff pair in, only not as followers into a bipolar diff pair. Otherwise very similar. Even the AGC gain adjust on the feedback.
In a new design the complexity would be replaced with a more straightforward ADC and fft. The tuning gets complicated with three stages and all the relays for fixed tuning with that much resolution. If it could be accomplished with the Maxim or JRC "pots" it might be something practical for a DIY project. otherwise the relay cost alone makes it impractical at best.
I looked at the Panasonic input amps and they are quite similar to the Shibasoku. JFET diff pair in, only not as followers into a bipolar diff pair. Otherwise very similar. Even the AGC gain adjust on the feedback.
If we used a synthesized sine pure in nature, in phase and amplitude matched with the input signal. A null process of removing the fundamental could be done.
Another approach could be to sample the input, clean up the signal and use that to null the signal to another sampler. Lots of things to play with here. The available technology is better these days.
It doesn't have to be a filtered sine it just has to be numerically correct. If applied to the input signal a little ahead of conversion then there enough time for settling before sampling. No filters required.
If Shibasoku can do this with 12 bits then what can we do with 16 bits.
Check out the LTC1609. I'm using it in my SVO. I picked up a surplus of 30 of them for a ridiculously low price.
They can withstand a +- 25Vp input. Ideal for what we are doing.
Another approach could be to sample the input, clean up the signal and use that to null the signal to another sampler. Lots of things to play with here. The available technology is better these days.
It doesn't have to be a filtered sine it just has to be numerically correct. If applied to the input signal a little ahead of conversion then there enough time for settling before sampling. No filters required.
If Shibasoku can do this with 12 bits then what can we do with 16 bits.
Check out the LTC1609. I'm using it in my SVO. I picked up a surplus of 30 of them for a ridiculously low price.
They can withstand a +- 25Vp input. Ideal for what we are doing.
Last edited:
Both are limited by their processes. matching the sine wave to null it would require phase locking a "perfect" sine wave to the incoming one and level matching. I can speculate on ways to do that but they probably would not get any better or less complex. (Cordell's distortion magnifier would be a practical implementation but will not create a perfect sine wave).
Sampling is limited to the capabilities of the ADC and DAC. The super oscillators we are dealing with are at least 10-20 dB better than the best ADC and DAC solutions today.Analog preprocessing does not need to be 3 cascaded notch filters. One with at least 20 db of notch depth would be enough to get on par with state of the art oscillators using available ADC's.
If we begin sampling at zero crossing with say 64 samples per cycle everything is predictable.
The 17th sample is peak. We can get the amplitude from that. Yes the input signal has to be phase locked and the sample clock generated from that. A sine look up table and a really good dac is all that's required. As you said with today's ADC's we can probably get away with a lot less than -140dB null.
An output frequency of 750Hz at 64 sample per cycle is a 48ksps. That give us 23.25kHz of bandwidth to view on a FFT app. Or that can be double to 1.5kHz at 96ksps and so on.
The 17th sample is peak. We can get the amplitude from that. Yes the input signal has to be phase locked and the sample clock generated from that. A sine look up table and a really good dac is all that's required. As you said with today's ADC's we can probably get away with a lot less than -140dB null.
An output frequency of 750Hz at 64 sample per cycle is a 48ksps. That give us 23.25kHz of bandwidth to view on a FFT app. Or that can be double to 1.5kHz at 96ksps and so on.
You have two options. First is to reconstruct the signal with a DAC and do analog analysis. The other would be to do FFT directly on the sampled data. Even an Arduino could do that today if you don't expect real time to 100 KHz. Display of the output becomes pretty easy and flexible. I suspect the later Panasonic analyzers work this way.